Field Evidence of the Effects of Privacy, Data Transparency, and Pro-social Appeals on COVID-19 App Attractiveness

Title of the Paper

Field Evidence of the Effects of Privacy, Data Transparency, and Pro-social Appeals on COVID-19 App Attractiveness

Paper Information

• Subject Area: Human-Computer Interaction and Privacy Research, focusing on adoption behaviors of COVID-19 contact tracing applications
• Keywords: COVID-19, privacy, data transparency, collective benefits, individual benefits, digital health, pro-social behavior, adoption behavior, gender differences, geographic differences

Research Background and Issues

• Issues or Challenges:

  1. User adoption rates of COVID-19 exposure notification applications are generally low, typically ranging between 10%-30%.
  2. Existing studies suggest that privacy concerns, lack of data transparency, and the framing of app promotion (emphasizing collective benefits vs. individual benefits) may influence adoption intentions.
  3. While laboratory studies and self-reported surveys have revealed user attitudes and tendencies, there is a lack of real-world ("field study") user behavior data.

• Research Importance:

  • High adoption rates of contact tracing apps can significantly improve the effectiveness of controlling the spread of COVID-19.
  • Understanding the specific impacts of privacy, transparency, and promotional framing on adoption behavior can help develop more targeted strategies for promoting health technologies.

• Research Motivation and Related Work:

  • This study is the first to evaluate the factors influencing the adoption of COVID-19 contact tracing apps in a real-world environment.
  • It complements existing research based on laboratory and self-reported data while providing data to support policymaking and app promotion.

Solution

• Research Methodology:

  • Collaborated with the state of Louisiana in the U.S. to conduct an experimental evaluation of adoption behavior through 14 randomly assigned advertising campaigns on Google Ads.
  • The advertisements were designed around three variables: privacy transparency, data transparency, and promotional framing (collective benefits vs. individual benefits).

• Experimental Design:

  1. Two promotional frames: "Collective benefits" (social benefits) vs. "Individual benefits" (personal benefits).
  2. Four types of privacy statements:
     • No privacy-related statement.
     • General privacy reassurance.
     • Non-technical privacy control ("You can control the data you share").
     • Technical privacy control ("Data stays on your device").
  3. Data transparency statements:
     • Indicating that the app will collect contact data or not specifying this.

• Experimental Procedure:

  • Advertisements were launched on the Google Ads platform, and user behavior in clicking on ads to navigate to the download page was recorded. Click-through rate (CTR) was used as the primary metric.
  • Demographic data (gender, age, geographic location) were collected to analyze the moderating effects of population factors.

Research Findings

• Specific Findings:

  1. Effects of Promotional Framing:
     • Ads emphasizing collective benefits ("Reduce COVID infections") had significantly higher click-through rates compared to those emphasizing individual benefits ("Receive exposure notifications").
  2. Effects of Privacy and Data Transparency:
     • Under the collective benefits framing, privacy statements (especially technical privacy) significantly increased click-through rates, but data transparency reduced click-through rates.
     • Under the individual benefits framing, technical privacy statements reduced click-through rates (with a stronger effect on men), while data transparency increased click-through rates.
  3. Impact of Demographics:
     • Gender: Women had higher overall click-through rates than men, especially under collective benefits ads.
     • Age: Older users (e.g., 65+) were more sensitive to ads, particularly those highlighting health risks relevant to them.
     • Geographic Location: Rural areas were more supportive of such app advertisements compared to urban areas.

• Advantages Over Existing Solutions:

  • Provides large-scale randomized controlled experimental data, supplementing evidence from self-reported and laboratory studies.
  • Demonstrates how to optimize messaging in real-world scenarios to encourage pro-social health behaviors.

• Experimental or Evaluation Results:

  • Click-through rates for collective benefits ads increased by approximately 34.1% to 45.8%.
  • Different combinations of privacy and data transparency statements had significant moderating effects on different user groups (gender and age).

• Limitations and Future Directions:

  • Unable to directly track app downloads and usage behavior following ad clicks.
  • The experiment was conducted in only one state (Louisiana), lacking cross-regional generalizability.
  • Limitations in the language design of the experiment; future studies could expand to include more linguistic styles and frameworks.
  • Future research is recommended to integrate technical behavior and ethical considerations while exploring best practices for further enhancing digital health behaviors.